According to Mark Cuban on Twitter, companies are intensifying efforts to safeguard their intellectual property (IP) and limit how AI models access and use proprietary training data, directly impacting the development of AI tools and synthetic data creation (source: Mark Cuban, Twitter, May 21, 2025). This trend could restrict the availability of high-quality training data for AI projects, potentially slowing innovation in blockchain and crypto startups that rely on AI-driven solutions. Traders should monitor regulatory and corporate actions regarding data access, as these could drive volatility in crypto assets tied to AI development.

According to Berkeley AI Research (@berkeley_ai), BAIR faculty members including Stuart Russell and Dan Klein discuss the future trajectory of large language models (LLMs) and the impact of synthetic data on AI development. Their insights focus on the potential changes in trading algorithms and market predictions driven by advancements in AI technology. The use of synthetic data is highlighted as a crucial factor in enhancing the accuracy and efficiency of trading systems.

According to @DeepLearningAI, researchers from Cohere have developed a method called active inheritance to optimize fine-tuning on synthetic data by selectively choosing training examples with desirable traits. This approach enhances model performance by reducing unwanted characteristics like toxicity, making it relevant for traders seeking reliable AI-generated insights.